U.S. patent application number 11/132730 was filed with the patent office on 2005-12-29 for aid device for inspection system and display method therefor.
Invention is credited to Hori, Masaki.
Application Number | 20050288881 11/132730 |
Document ID | / |
Family ID | 34936662 |
Filed Date | 2005-12-29 |
United States Patent
Application |
20050288881 |
Kind Code |
A1 |
Hori, Masaki |
December 29, 2005 |
Aid device for inspection system and display method therefor
Abstract
An aid device is set to an inspection system for obtaining
waveform data from an inspection object, calculating a value of a
characteristic quantity that characterizes the inspection object
and making a judgment whether the inspection object is normal or
abnormal. The aid device selects one or more characteristic
quantities and one or more parameters, calculates a value by using
specified combinations of the selected characteristic quantities
and parameters to obtain calculation results, and displays a graph
based on each of the calculation results for the specified
combinations. The graph has at least two axes, each representing a
characteristic quantity set or a parameter set and makes a display
by varying density, color, height or size corresponding to the
results of the calculation.
Inventors: |
Hori, Masaki; (Yokohama,
JP) |
Correspondence
Address: |
BEYER WEAVER & THOMAS LLP
P.O. BOX 70250
OAKLAND
CA
94612-0250
US
|
Family ID: |
34936662 |
Appl. No.: |
11/132730 |
Filed: |
May 18, 2005 |
Current U.S.
Class: |
702/82 |
Current CPC
Class: |
G01N 29/42 20130101;
G01D 1/18 20130101; G01N 29/14 20130101; G01N 29/4427 20130101;
G01D 1/14 20130101; G01N 29/46 20130101 |
Class at
Publication: |
702/082 |
International
Class: |
G06F 019/00 |
Foreign Application Data
Date |
Code |
Application Number |
May 18, 2004 |
JP |
P2004-148445 |
May 16, 2005 |
JP |
2005-143321 |
Claims
What is claimed is:
1. An aid device for an inspection system, said inspection system
serving to obtain waveform data from an inspection object, to
calculate a value of a characteristic quantity that characterizes
said inspection object and to thereby make a judgment whether said
inspection object is normal or abnormal, said aid device
comprising: selecting means for selecting one or more
characteristic quantities and one or more parameters; calculating
means for calculating a value for given waveform data by using
specified combinations of the selected characteristic quantities
and parameters; and display means for displaying a graph based on
results of calculation by said calculating means, said graph having
at least two axes, each of said two axes being formed by one
selected from the group consisting of said one or more
characteristic quantities and said one or more parameters selected
by said selecting means; wherein said aid device serves to
determine at least either of an effective characteristic quantity
for said judgment and a parameter for calculating a value of said
effective characteristic quantity.
2. The aid device of claim 1 wherein said waveform data consist of
two object waveform data for comparison; wherein said calculating
means calculates values specified characteristic quantities for
said two object waveform data by using said specified combinations
to obtain calculation results and carries out comparison
calculation between said calculation results; and wherein said
graph displayed by said display means is based on results of said
comparison calculation.
3. The aid device of claim 2 wherein said graph makes a display by
varying one selected from the group consisting of density, color,
height and size corresponding to the results of said comparison
calculation.
4. The aid device of claim 1 wherein said waveform data consist of
a plurality of waveform data belonging to a same group; and wherein
said calculating means calculates values for each of said plurality
of waveform data by using said specified combinations to thereby
obtain calculation results and determines at least either of an
effective characteristic quantity for said judgment and a parameter
for calculating value of effective characteristic quantity based on
the calculation results on each of said specified combinations for
said plurality of data.
5. The aid device of claim 4 wherein said graph makes a display by
varying one selected from the group consisting of density, color,
height and size corresponding to the results of said comparison
calculation.
6. A display method for an aid device for an inspection system,
said inspection system serving to obtain waveform data from an
inspection object, to calculate a value of a characteristic
quantity that characterizes said inspection object and to thereby
make a judgment whether said inspection object is normal or
abnormal, said aid device serving to determine at least either of
an effective characteristic quantity for said judgment and a
parameter for calculating value of said effective characteristic
quantity, said method comprising the steps of: selecting one or
more characteristic quantities and one or more parameters;
calculating a value by using specified combinations of the selected
characteristic quantities and parameters to thereby obtain
calculation results; and displaying a graph based on each of said
calculation results for said specified combinations, said graph
having at least two axes, each of said two axes being formed by one
selected from said selected characteristic quantities and
parameters.
7. The method of claim 6 wherein said graph makes a display by
varying one selected from the group consisting of density, color,
height and size corresponding to the results of said comparison
calculation.
Description
[0001] Priority is claimed on Japanese Patent Applications
2004-148445 filed May 18, 2004 and 2005-143321 filed May 16,
2005.
BACKGROUND OF THE INVENTION
[0002] This invention relates to an aid device for an inspection
system such as an abnormal sound detecting system and a method of
display for such a device.
[0003] Very many rotary machines incorporating a motor are used in
automobiles and household electric appliances. Rotary machines
mounted to an automobile include the engine, the power steering
machine, the power seats and the transmission. Household electric
appliances incorporating a motor include refrigerators, air
conditioners and washing machines. When such a rotary machine is
operated, a sound is generated due to the rotation of the
motor.
[0004] Sounds that originate from a motor include both a normal
sound which is necessarily generated with the rotation of the motor
and an unusual sound generated when something is wrong with the
motor. An unusual sound may be generated for different causes such
as an abnormality in the bearing, an abnormal contact inside, an
unbalanced condition and the introduction of a foreign object. If
an unusual sound occurs at a regular rate of once per rotation, the
possible cause may be a chipped gear, a foreign object that may
have been caught inside, a spot defect and the instantaneous
rubbing between a rotary component and a stationary component.
Sounds that give an unpleasant sensation may be of any frequency
within the audible range between 20 Hz and 20 kHz. An unpleasant
sound may have a frequency of about 15 kHz. In other words, a sound
of a specified frequency component can be an abnormal sound. It
naturally goes without saying that abnormal sounds can be of any
frequency.
[0005] Sounds that accompany an abnormal condition are unpleasant
but may also cause another kind of abnormal condition. Thus, it is
a common practice at a production factory to carry out a sensory
test, relying upon the five senses of inspectors in order to detect
abnormal conditions and this includes listening with ears and
touching by the hands to feel vibrations. A sensory test may mean
any test conducted by using the human ability to sense.
[0006] In the technical field of automobile production, there has
recently been a significant increase in the requirement regarding
the sound quality, that is, the need for an automated quantitative
test on moving parts of automobiles such as the engine and the
transmission. Qualitative tests of the conventional type such as
sensory tests as described above can no long satisfy the needs of
the modern days.
[0007] In view of this situation, devices for detecting abnormal
sounds are coming to be developed for reliably carrying out a
quantitative test according to a clear standard. These are devices
for automating a conventional sensory test, adapted to measure the
vibrations and sound from a moving part of a product and to inspect
the frequency components of analog signals by means of a frequency
analyzer making use of an FFT algorithm or the like, as described
in Japanese Patent Publication Tokkai 11-173909, although a
bandpass filter may be used instead for the analysis of analog
signals.
[0008] A frequency analyzer, as disclosed in aforementioned
Japanese Patent Publication Tokkai 11-173909, is adapted to analyze
a time signal into a frequency region by means of a fast Fourier
transform algorithm. Since the frequency range of abnormal sounds
is more or less determined, it is possible to extract frequency
components corresponding to a specified range and to obtain a
characteristic quantity from the extracted component. Thus, the
absence and presence of an abnormal condition and the cause of its
presence can be estimated from such a characteristic quantity by
solving a fuzzy logic problem or the like. Such an inspection
system can not only make determinations automatically according to
a once determined standard but also record and store the waveform
data at the time of the determination in a memory device within the
system.
[0009] Such an inspection system for an abnormal sound is required
to select various parameters to be used for selecting optimum
characteristic quantities for the inspection and calculating values
of the selected characteristic quantities. At the present time,
however, the selection of characteristic quantities and parameters
is being made by the user based only on his/her experience and
instinct. Although Japanese Patent Publication Tokkai 9-44465 on
optimization method and device using a genetic (hereditary)
algorithm disclosed hierarchical genetic and parallel genetic
algorithms that may be expected to improve the accuracy of
inspection, conventional inspection systems such as disclosed in
aforementioned Japanese Patent Publication Tokkai 11-173909 still
rely upon human experience and instinct in the extraction of
characteristic quantities and selection of parameters for the
calculation of the characteristic quantities. This means that human
experience and instinct are to be relied upon to an unreasonable
degree in order to determine the presence or absence of an abnormal
condition and to select parameters for the calculation of
characteristic quantities for such a determination based on
thousands of data items.
[0010] When a waveform is analyzed, for example, the given waveform
is characterized in various ways. In order to obtain one such
characteristic quantity, usually a number of parameters must be
used such that the value of the characteristic quantity will change
as these parameters are varied. If parameters are selected
appropriately, the characteristic of the given waveform becomes
more apparent as the value of this characteristic quantity. Thus,
it is extremely important to adjust these parameters.
[0011] Even with a single parameter, there are many patterns to be
set. Thus, it is a complicated work to compare the calculated
results of the characteristic quantity while the setting is
sequentially changed and it is difficult to set the parameter
correctly. It is also difficult to ascertain in which of the
plurality of characteristic quantities the characteristics of the
waveform will appear as the number of combinations of the
parameters increases.
[0012] In the case of a production line, in particular, an early
determination of optimum parameters is very important. If human
experience and instinct alone are relied upon, the determination is
too time-consuming.
[0013] Problems also exist in an attempt to apply a hierarchical
genetic algorithm of aforementioned Japanese Patent Publication
Tokkai 9-44465 for the determination of optimum parameters for an
abnormal sound inspection system. Parameters must be set by a
trial-and-error method even for a genetic algorithm without a
hierarchical structure. For accumulating such parameters in a
hierarchical structure, a trial-and-error routine just as
cumbersome as the selection of a characteristic quantity and a
parameter by experience and instinct would be required in order to
obtain a desired result.
[0014] Since the control of a genetic algorithm itself becomes
complicated, it is difficult to incorporate a search method
corresponding to the degree of influence among the parameters.
Thus, even if a method of Japanese Patent Publication Tokkai
9-44465 is used, it is difficult to obtain optimum parameters
effectively in a short time.
[0015] In the case of a genetic algorithm, furthermore, even if
characteristic quantities or parameters to be finally used by an
abnormal sound inspection system are determined, it is not possible
for the user to learn how they were determined or to find out if
they are really optimum quantities or parameters. Since an accurate
judgment cannot be made without the conditions being optimized, the
inspection of abnormal sound will have to be carried out by blindly
believing that the determined characteristic quantities or
parameters are optimum ones.
SUMMARY OF THE INVENTION
[0016] It is therefore an object of this invention to provide an
aid device and a display method for an inspection system for
obtaining waveform data from an inspection object, calculating a
value of a characteristic quantity that characterizes the
inspection object and thereby making a judgment whether the
inspection object is normal or abnormal, by aiding the user to
effectively determine characteristic quantities and parameters and
to visually ascertain that they are optimum as well as the
difference between a normal waveform and an abnormal waveform.
[0017] An aid device embodying this invention may be characterized
as comprising selecting means for selecting one or more
characteristic quantities and one or more parameters, calculating
means for calculating a value for given waveform data by using
specified combinations of the selected characteristic quantities
and parameters, and display means for displaying a graph based on
results of calculation by the calculating means, the graph having
at least two axes, each of these two axes being formed by one
selected from the group consisting of the one or more
characteristic quantities and one or more parameters selected by
the selecting means.
[0018] In the above, the waveform data may consist of two object
waveform data for comparison and the calculating means may be
adapted to calculate the values for these two object waveform data
by using the specified combinations to obtain calculation results
and to carry out comparison calculation between the calculation
results, the graph displayed by the display means being based on
results of this comparison calculation. The graph may be arranged
to make a display by varying density, color, height or size
corresponding to the results of the comparison calculation.
[0019] The waveform data may consist of a plurality of waveform
data belonging to a same group and the calculating means may be
adapted to calculate the value for each of these plurality of
waveform data by using the specified combinations to thereby obtain
calculation results and to determine at least either of an
effective characteristic quantity for the judgment and a parameter
for calculating a value of the effective characteristic quantity
based on the calculation results on each of the specified
combinations for the plurality of data.
[0020] A display method embodying this invention is for an aid
device as described above and may be characterized as comprising
the steps of selecting one or more characteristic quantities and
one or more parameters, calculating a value by using specified
combinations of the selected characteristic quantities and
parameters to thereby obtain calculation results and displaying a
graph, as described above, based on each of the calculation results
for the specified combinations, the graph having at least two axes,
each of these two axes being formed by one selected from the
selected characteristic quantities and parameters.
[0021] According to this invention, a graph is displayed by using
characteristic quantities and parameters to form two or more
coordinate axes, corresponding to the calculated results. Thus, the
user can easily determine visually which combinations of
characteristic quantities and parameters are effective and hence
can easily grasp the characteristics of given waveform data, that
is, the difference among a plurality of waveform data.
[0022] Throughout herein, the expression "parameter" is intended to
mean any item that may be set in and can influence the calculation
of a value of a characteristic quantity for given waveform data.
The following two kinds of such parameter may be considered with
reference to a given characteristic quantity. Parameters of one
kind are those for a preliminary process to be carried out on a
measured waveform before a value of the characteristic quantity is
actually calculated. If such parameters are changed, the waveform
that is inputted to the calculating means for carrying out the
calculation will change even if the measured waveform is the same.
Examples of parameter of this kind include constants related to a
filter. If the upper and lower limit values of a frequency filter
are changed, for example, the change affects the frequency
components that pass through the filter and hence also the waveform
that is inputted to the calculating means for calculating a value
of its characteristic quantity. Other examples of a parameter for
such preliminary processing include the number of smoothing data
for an envelop line in a waveform conversion process. Parameters of
the other kind are for the calculation of a value of a
characteristic quantity itself, that is, parameters that are
essentially required for carrying out the calculation, affecting
the results of the calculation even if the waveform inputted to the
calculating means is the same. Examples of parameter of this kind
include those for specifying a target frequency range for the
characteristic quantity or threshold values with which
characteristic quantities are compared.
BRIEF DESCRIPTION OF THE DRAWINGS
[0023] FIG. 1 is a block diagram of an aid device embodying this
invention.
[0024] FIG. 2 is an example of display by the waveform data input
indicating part.
[0025] FIG. 3A is an example of waveform data reading screen, and
FIG. 3B is an example of waveform data saving screen.
[0026] FIG. 4 is an example of calculation mode selection screen of
the calculation mode switching part.
[0027] FIG. 5 is an example of selection screen by the axis
selecting part.
[0028] FIGS. 6-8 are examples of setting screen by the axis pattern
setting part.
[0029] FIGS. 9 and 10 are examples of display screen by the
calculation result display part.
[0030] FIG. 11 is a drawing for showing the principle of comparison
calculation carried out by the calculating part.
[0031] FIG. 12 is an example of combined display screen.
[0032] FIG. 13 is a functional block diagram of a portion including
the calculating part according to a first embodiment of the
invention.
[0033] FIGS. 14 and 15 are a flowchart for explaining the functions
of the first embodiment of the invention.
[0034] FIG. 16 is a functional block diagram of a portion including
the calculating part according to a second embodiment of the
invention.
[0035] FIG. 17 is a flowchart for explaining the functions of the
second embodiment of the invention.
[0036] FIGS. 18 and 19 are a flowchart for explaining a portion of
the functions of a third embodiment of the invention.
[0037] FIG. 20 is a functional block diagram of a portion including
the calculating part according to a fourth embodiment of the
invention.
[0038] FIG. 21 is a functional block diagram of a portion including
the calculating part according to a fifth embodiment of the
invention.
[0039] FIGS. 22 and 23 are a flowchart of the operations carried
out by an aid device according to the fifth embodiment of the
invention.
[0040] FIG. 24 is a functional block diagram of a portion including
the calculating part according to a sixth embodiment of the
invention.
[0041] FIGS. 25 and 26 are a flowchart for explaining the functions
of the sixth embodiment of the invention.
[0042] FIG. 27 is a flowchart for an example of using this
embodiment.
DETAILED DESCRIPTION OF THE INVENTION
[0043] An abnormal sound inspection system for which characteristic
quantities and parameters are set according to the present
invention will be briefly explained first. This inspection system
is adapted to carry out a preprocessing on waveform data obtained
through a vibration sensor or a sound microphone, thereafter to
calculate values of specified characteristic quantities and to
judge the object of inspection to be "good", "no good" or
"undetermined" by using effective ones of the calculation results.
Different kinds of filters such as bandpass filters, low pass
filters and high pass filters are provided for the preprocessing,
and many kinds of characteristic quantities are also prepared.
[0044] For each object of inspection, the types of preprocessing
and characteristic quantities that are effective in determining
whether it is "good" or "no good" are sometimes more or less known
such that it turns out to be wasteful to calculate values of
characteristic quantities that are not very effective. According to
this invention, therefore, a function is provided for providing
data for determining which characteristic quantities are suitable
for the object of inspection. Although a method of calculation is
determined for each characteristic quantity, its value obtained by
the calculation and hence any judgment based thereupon also will
change as parameters are changed. In other words, even if an
originally effective characteristic quantity is used, a wrong
judgment may be rendered unless parameters are correctly set. An
aid device according to this invention is therefore provided with a
function of providing data for easily finding suitable combinations
of characteristic quantities and parameters. Moreover, such
provided data are easily understandable visually, displayed on a
display device in color or in a three-dimensional representation.
For allowing an accurate judgment, numeral data are also
displayed.
[0045] Explained more in detail, an aid device according to this
invention is adapted to calculate a value on the basis of
preliminarily set conditions on characteristic quantities and
parameters as waveform data for one or more samples are received
and to obtain the value of the result of this calculation or its
normalized value (hereinafter referred to as the evaluation value).
Data based on the evaluation value thus obtained are displayed by
using coordinate axes on a display of the aid device. This may be
done by assigning the characteristic quantity set and parameter set
to the vertical and horizontal axes before the display is made, and
the aid device calculates a value under a combination of
characteristic quantity and parameter on the vertical and
horizontal axes to make a display on each coordinate based on the
calculated evaluation value. Since each axis can represent either a
characteristic quantity set or a parameter set, combinations
between two characteristic quantities, between a characteristic
quantity and a parameter and between two parameters are all
possible. When the combination is between two parameters, a
characteristic quantity is separately set.
[0046] FIG. 1 shows a preferred embodiment of the invention with an
aid device 10 provided with a waveform data input indicating part
11, a calculation mode switching part 12, an axis selecting part 13
and an axis pattern setting part 14 as an input interface with the
user. The output interface of the aid device 10 is provided with a
calculation result display part 15 for displaying results of
calculations carried out by a calculating part 20. According to the
present embodiment, the input interface and the output interface
are realized by displaying input/output screens with specified
layouts on the display screen of a display device.
[0047] The aid device 10 further comprises the calculating part 20
for calculating values of characteristic quantities according to
commands from the input interface, a waveform data input part 16
for providing the calculating part 20 with waveform data according
to commands from the waveform data input indicating part 11, a
memory device 17 for storing waveform data and an A/D sampling part
18 for obtaining waveform data. The A/D sampling part 18 is
connected to a waveform signal detecting means such as a vibration
sensor or a sound microphone to sample analog waveform signals
received from the connected waveform signal detecting means and to
generate digital waveform data.
[0048] The waveform data input indicating part 11 is for indicating
waveform data to be provided to the calculating part 20 in order to
determine conditions for characteristic quantities and parameters.
The waveform data to be provided may be existing data stored in the
memory device 17 or data that were actually sampled and collected.
Which data should be used is indicated by the waveform data input
indicating part 11. If the waveform data stored in the memory
device 17 are to be used, the file containing the data to be used
is indicated. The number of data to be indicated may be only one or
may be more than one. Waveform data may be inputted to the memory
device 17 as a group. For example, waveform data belonging to a
group of "good" products, those belonging to a group of "no good"
products and those belonging to a specified kind of products may be
inputted.
[0049] The waveform data input part 16 serves to receive the
specified waveform data according to a received command and to
deliver them to the calculating part 20. In other words, when a
file name is received, the waveform data input part 16 accesses the
memory device 17 to retrieve the waveform data of the received file
name and transmits them to the calculating part 20. If the sampling
of waveform data is specified, a sampling command is given to the
A/D sampling part 18 to obtain the sampling data provided therefrom
and to transmit them to the calculating part 20 as waveform
data.
[0050] FIG. 2 shows an example of input indicating screen of the
waveform data input indicating part 11 displayed on the screen of
the display device. In this example, two input indicating areas
"Waveform Input 1" and "Waveform Input 2" are prepared. These two
input indicating areas have the same functions. It is so arranged
that waveform data that are inputted by using the same waveform
input area form the same group. Thus, for example, the difference
between the characteristic quantities of waveform data belonging to
two different groups such as groups of "good" and "no good"
products can be easily grasped. Thus, these areas can be used such
that if waveform data of "good" products are obtained by using
input indicating area "Waveform Input 1", they will belong to a
first group and if waveform data of "no good" products are obtained
by using input indicating area "Waveform Input 2", they will belong
to a second group. The input indicating areas can thus be used such
that the groups of waveform data may be specified.
[0051] Each input indicating area is provided with button areas,
11b and 11c for inputting various commands. For example, if the
"Collect Data" button 11a is clicked, sampling of waveform data is
indicated to the waveform data input part 16. If the "Read File"
button 11b is clicked, a waveform data reading screen as shown in
FIG. 3A is displayed to request that a file to be read be
specified. When a file name is specified on this waveform data
reading screen, the address of storage and the file name are
transmitted to the waveform data input part 16.
[0052] Each input indicating area is also provided with a waveform
data display area 11d for displaying waveform data obtained by the
waveform data input part 16 such as those obtained from the A/D
sampling part 18 and sampled as well as those read out of the
memory device 17. The waveform data displayed in the waveform data
display area 11d can be stored in the memory device 17 by clicking
the "Save File" button 11c. When the "Save File" button 11c is
clicked, the waveform data saving screen as shown in FIG. 3B is
displayed. As the column on the left-hand side of this waveform
data saving screen is used to specify a holder for saving and a
file name is written in, the waveform data displayed in the
waveform data display area 1d are stored at the specified address.
This function is usually used for storing waveform data collected
by sampling in the memory device 17.
[0053] On the right-hand side of the waveform data display area
11d, there is a waveform list display area 11e for displaying a
list of waveform data that have been inputted by using the input
indicating areas. Unwanted waveform data can be deleted from the
list displayed in this waveform list display area 11e by selecting
them on this list and then clicking a "Delete" button 11f.
[0054] These two input indicating areas are not always used in all
applications. They may be used, for example, for comparing a
plurality of different waveforms. When conditions of a
characteristic quantity or a parameter are being obtained based on
a single waveform data item or a plurality of waveform data
belonging to a single group, use may be made of only one of the
input indicating areas.
[0055] The calculation mode switching part 12 is for selecting the
contents of calculation and display. As explained above, the
waveform data used in the calculations are divided into two groups.
The waveform data to be provided to the calculating part 20 may be
only those belonging to one of the groups or those belong to both
groups. When a value of a characteristic quantity is calculated
based on waveform data belonging to both groups, the results of
calculations may be further compared such that the difference of
characteristics of the plurality of waveforms can be more easily
grasped. For this purpose, as waveform data belonging to two groups
are inputted, the calculation mode switching part 12 serves to
specify the type of the comparison calculations and the selected
calculation mode is provided to the calculating part 20.
[0056] FIG. 4 shows an example of calculation mode selection screen
of the calculation mode switching part 12 displayed on the display
screen of the display device. By this example, as shown, there are
three modes "Waveform 1", "Waveform 2" and "Comparison" that can be
selected. If "Waveform 1" is selected, a value of a characteristic
quantity is calculated based on waveform data inputted by using
"Waveform Input 1". If "Waveform 2" is selected, a value of a
characteristic quantity is calculated based on waveform data
inputted by using "Waveform Input 2". If "Comparison" is selected,
values of a characteristic quantity are calculated based on
waveform data of two inputted groups by using both "Waveform Input
1" and "Waveform Input 2". In this case, "Comparison Calculation
Content" is selected from the list of comparison calculations
prepared in the pull-down menu format.
[0057] The axis selecting part 13 is for setting the vertical and
horizontal axes of the rectangular coordinate system in the graph
for displaying the results of the calculations. FIG. 5 shows an
example of selection screen by the axis selecting part 13 displayed
on the display screen of the display device. Either a
characteristic quantity or a parameter may be selected each as the
vertical axis or as the horizontal axis. When two parameters are
selected for the two axes for a single characteristic quantity,
this characteristic quantity is indicated in the first line. A
proper characteristic quantity may be selected from a list prepared
similarly in a pull-down menu format. If a characteristic quantity
is selected for either of the axes, "none" is indicated on this
line, as shown as an example in FIG. 5.
[0058] On the second line, either a parameter or a characteristic
quantity is selected for each of the axes. For each of the
selections, a detailed list is prepared and a detailed item is
selected. If parameters corresponding to the same detailed item are
selected for both axes, the calculation formula to be carried out
between the axes is indicated in the third line by selecting from a
preliminarily prepared list. The selections thus made are
transmitted to the calculating part 20 and required data are sent
to the axis pattern setting part 14.
[0059] The axis pattern setting part 14 is for actually setting up
the data assigned to each of the axes. If a characteristic quantity
is indicated, for example, all selectable candidates are displayed
such that the one to be calculated is selected therefrom and the
selected characteristic quantity is set to the associated axis.
[0060] These item data specified through the axis selecting part 13
are transmitted not only to the calculating part 20 but also to the
axis pattern setting part 14. The axis pattern setting part 14
serves to actually set up the detailed contents of the
characteristic quantity and/or parameter selected for each axis
through the axis selecting part 13 and to provide them to the
calculating part 20. In the case of an axis representing a
characteristic quantity, for example, a selection is made to
determine which of the characteristic quantities should be used. In
the case of an axis representing a parameter, detailed numerical
values for the parameter selected as the detailed item may be set
up.
[0061] FIGS. 6-8 show examples of setting screen by the axis
pattern setting part 14 displayed on the display screen of the
display device. The screen shown in FIG. 6 is for setting up a
characteristic quantity set by displaying a list of usable
characteristic quantities. The user may operate a pointing device
to click inside the square symbol in front of the name of the
selected characteristic quantity to thereby mark the square. The
prepared characteristic quantities may include RMS (Root Mean
Square) and maximum amplitude level. It goes without saying that
the selection as shown in FIG. 6 is not intended to limit the scope
of the invention. Such selection may be dispensable such that all
characteristic quantities may become the object of
calculations.
[0062] FIG. 7 shows an example where a bandpass filter is selected
as parameter with 31 pass band ranges settable as parameter. As
shown, each pass band is indicated by its lower and upper limit
values. An initial value of zero may be set in all these areas such
that the user is required to fill them. Since frequency ranges
where a characteristic is likely to appear are more or less
anticipated, however, they may be preliminarily set as sets of
initial values. Unlike the example shown in FIG. 7, however, one
parameter may be set from zero to infinity such that a
characteristic quantity not dependent upon frequency may be
obtained.
[0063] FIG. 8 is an example of setting screen where up to 31 kinds
of data parts to be used (such as parts of data that are used where
the amplitude is particularly large) are settable as parameters. In
this example, too, appropriate initial values may be displayed such
that the user may be allowed to change them if necessary. This
example may be used, for example, where the maximum amplitude level
is selected as the characteristic quantity and a specified number
(N) of used data part from the largest amplitude are to be
extracted as characteristic quantities.
[0064] Buttons marked "Read" and "Save" may be provided, as shown
in FIGS. 6-8, to be used for saving the patterns of characteristic
quantities and parameters that have been created and reading out a
pattern created and saved in the past to be used again. These
patterns may be stored in the memory device 17 or another separate
memory means. A detachable recording medium may be used to store
patterns of model characteristic quantities and parameters such
that they can be retrieved and used.
[0065] The calculating part 20 is for carrying out calculation
processes on waveform data provided from the waveform data input
part 16 according to conditions set by the input interface of
various types described above. The results of these calculation
processes are transmitted to the calculation result display part
15. The structure of the calculating part 20 will be explained
below in more detail.
[0066] The calculation result display part 15 serves to create a
calculation result display screen including a result display graph,
based on the calculation results provided from the calculating part
20, and to display it in a specified area on the display device.
FIG. 9 shows an example of calculation result display screen.
Display graph G1 shows a rectangular coordinate system with its
vertical and horizontal axes representing the selected
characteristic quantity set or parameter set, the values set by the
axis pattern setting part 14 being set along each axis. Each
quadrangular area in the graph is shown by a level of density
corresponding to the result of calculating a value of a
characteristic quantity under the conditions associated with the
corresponding vertical and horizontal axes or a normalized result
obtained therefrom (the evaluation value). In the illustrated
example, the higher the evaluation value, the higher the level of
density of the corresponding area. The evaluation value is
classified into several levels (5 levels according to the example)
and the area is shown by a black color if the evaluation value
there belongs to the highest level. The color is white if the
evaluation value belongs to the lowest level.
[0067] With the evaluation values thus represented by the density
of areas, the user can easily locate the areas with a high level of
density and hence the combination of corresponding characteristic
quantity and parameter. In other words, a combination of
characteristic quantity and parameter that is effective to the
object waveform data can be easily selected.
[0068] According to the illustrated example, there are hair cursors
C1 and C2 provided along the vertical and horizontal direction,
movable, say, by using a pointing device and dragging them after
placing its pointer on them. Both cursors may be simultaneously
moved by dragging if the pointer is placed at the crossing point of
the two cursors C1 and C2.
[0069] A vertical-axis graph G2 and a horizontal-axis graph G3 are
also outputted and displayed respectively on the right-hand side
and below the main display graph G1, respectively displaying the
evaluation values of the areas on the hair cursors C1 and C2. The
vertical-axis and horizontal-axis graphs G2 and G3 show evaluation
values not only by the density but also by height such that the
user can more easily locate areas with a high evaluation value. The
height of these bar graphs may be arranged such that areas with the
same level of density will have the same height or the height will
vary according to the actual evaluation value even if two areas
show a same level of density.
[0070] A value display area R may be further provided for
displaying the actual numerical value of the evaluation value of
the area indicated by the two hair cursors C1 and C2. From the
numerical outputs thus made, the user can more reliably select a
combination of characteristic quantity and parameter that provides
a higher evaluation value.
[0071] FIG. 9 shows an example where the vertical axis represents
characteristic quantities and the horizontal axis represents
parameters, showing that the combination of Characteristic Quantity
D and Parameter 4 results in a high evaluation value and suggesting
that this combination may be used as one of candidates to be used
as condition for actually carrying out a calculation of
characteristic quantity for an abnormal sound detecting system. The
example of FIG. 9 shows, however, that there are two areas where
the evaluation value becomes very high. In this situation, the hair
cursors C1 and C2 may be moved as shown in FIG. 10 such that a
comparison can be made between the actual numerical evaluation
values.
[0072] It is preferable that calculated results of characteristic
quantities be normalized. This is because the maximum value of
calculation varies frequently, depending on the kind of the
characteristic quantity. In order to compare evaluation values for
different characteristic quantities, therefore, more accurate
judgment can be made by normalizing the evaluation values such that
the values can be compared on the same scale.
[0073] Although the magnitude of evaluation values for a
characteristic quantity is represented by the level of density,
this is intended to be merely an example. Different colors or
circles with different radii may be used to distinguish among
different levels of evaluation values. Even numerical values may be
directly written in on the graph. What is essential according to
this invention is that a display be made such that the user can
visually distinguish a favorable combination corresponding to a
high evaluation value.
[0074] FIG. 11 shows an example where the results of calculating
characteristic quantities based on waveform data belonging to two
different groups are being compared by calculation. After values of
characteristic quantities are calculated based on waveform data
belonging to each of two groups and graphs like the main graph G1
shown in FIG. 9 are prepared, a comparison process is carried out,
say, by calculating the difference between these two graphs to
finally obtain a comparison result graph. Since combinations for
which the difference in evaluation values is large between the
groups can thus be clearly picked out, preferable characteristic
quantities and parameters for distinguishing between the groups can
be easily obtained. In the case of the illustrated example, the
combination of Characteristic Quantity D and Parameter 4 provides a
large evaluation value but since a large evaluation value is
obtained by waveform data belonging to either of the groups, it may
be concluded that it is not a combination suitable for
distinguishing between these two groups.
[0075] FIG. 12 shows an example of display screen on the display
device, unifying the individual display screens of the
aforementioned input and output interfaces. According to this
example, two input indicating areas for waveform data are prepared
at the top, each input indicating area being provided with two
waveform display areas of which the one on the left-hand side is
for showing the whole of a waveform and the other on the right-hand
side is for showing a portion thereof in an enlarged form along the
time-axis in the horizontal direction. The area for displaying a
list may be positioned in any appropriate manner. The screen for
setting a calculation mode shown in FIG. 4 corresponds to the
"object of analysis" at the center of FIG. 12. The layout of the
"Comparison" button, etc. is also modified. If the sheet with the
heading "BPF" shown in FIG. 12 is selected, a pattern setting
screen as shown in FIG. 7 is displayed. Since the example shown in
FIG. 12 is adapted to obtain all of the prepared characteristic
quantities, there is no setting screen for characteristic
quantities as shown in FIG. 8. If a selection is to be made, such a
display may be made superposed to this sheet. The bottom left-hand
side of FIG. 12 is for displaying a graph of the calculation
results, etc.
[0076] Functions of the calculating part 20 are explained next more
in detail. FIG. 13 shows the calculating part 20 operating
according to a first embodiment of the invention. Characteristic
quantities and parameters are set respectively along the vertical
and horizontal axes by means of the axis selecting part 13 and
either "Waveform 1" or "Waveform 2" is selected by the calculation
mode switching part 12. One waveform data item is obtained by using
the corresponding input indicating area and this waveform data item
is provided to the calculating part 20 through the waveform data
input part 16. The calculating part 20 includes a sequential
calculating part 21 for calculating values of characteristic
quantities sequentially according to a set condition.
[0077] FIG. 14 is a flowchart of the sequence of processes carried
out by the calculating part 20 according to the first embodiment.
To start, the sets of parameters and characteristic quantities set
by the axis selecting part 13 and the axis pattern setting part 14
are read in (Step S1). Examples of the set of characteristic
quantities include RMS and maximum amplitude level. Examples of the
parameter set include combinations of upper and lower limit values
of a band pass filter such as 13 Hz-18 Hz, 20 Hz-28 Hz and 25 Hz-35
Hz. Next, the waveform data which are the object of processing are
read in from the waveform data input part 16 (Step S2). The order
in which Steps S1 and S2 are carried out may be reversed.
[0078] Next, values are calculated for combinations of all
parameters and all characteristic quantities that were set by
executing Step S1 for given waveform data (Step S3). If the
parameters are the pass bands of a band pass filter as described
above, for example, the calculations of values of characteristic
quantities are carried out for the frequency components that have
passed for the waveform data read in by executing Step S2.
Evaluation values are normalized for each characteristic quantity
by using the maximum evaluation value such that parameters that are
effective for each characteristic quantity can be easily
identified.
[0079] The evaluation values obtained by executing Step S3 are
transmitted to the calculation result display part 15 such that the
calculation results are displayed all together (Step S4). If the
horizontal axis is set for parameters and the vertical axis is set
for characteristic quantities by the axis selecting part 13, for
example, a display may be made as shown in FIG. 9 with parameters
arranged along the horizontal axis, characteristic quantities
arranged along the vertical axis and the magnitudes of the
evaluation values each calculated by combining a parameter and a
characteristic quantity shown by different densities. As explained
above, the display may be made by using different colors or as a
three-dimensional graph. What is important is that the display be
made such that the user can easily ascertain visually where to find
a high evaluation value on the graph for each characteristic
quantity for the waveform data that have been read in.
[0080] Step S3 of FIG. 14 is carried out by the aforementioned
sequential calculating part 21 according to the flowchart of FIG.
15. To start, the first of the set of the parameters is selected
(Step S11) and the first of the set of characteristic quantities is
selected (Step S12) such that one of the conditions is set each
along the vertical and horizontal axes.
[0081] Next, the parameter that has been selected is used to
calculate the value of the characteristic quantity that has been
selected for the obtained waveform data (Step S13). This is a
similar calculation process carried out, for example, in an
ordinary abnormal sound inspection system. It is determined
thereafter if the last of the characteristic quantities in the set
has been selected (Step S14) and if the process of Step S13 was not
for the last characteristic quantity (NO in Step S114), Step S13 is
repeated after the next characteristic quantity in the set is
selected (Step S15). If the process of Step S13 was for the last of
the set of the characteristic quantities (YES in Step S14), it is
determined if the last of the parameters in the parameter set has
been selected (Step S16) and if it was not the last parameter (NO
in Step S16), Step S12 is repeated after the next parameter is
selected (Step S117). This is repeated until the last of the
parameters in the set is selected (Yes in Step S16) and Step S3 in
the flowchart of FIG. 14 is completed.
[0082] FIG. 16 shows the calculating part 20 operating according to
a second embodiment of the invention. Parameters are set along both
the vertical and horizontal axes by means of the axis selecting
part 13 and either "Waveform 1" or "Waveform 2" is selected by the
calculation mode switching part 12. One waveform data item is
obtained by using the corresponding input indicating area and this
waveform data item is provided to the calculating part 20 through
the waveform data input part 16. The calculating part 20 includes a
sequential calculating part 21 for calculating values of
characteristic quantities sequentially according to a set
condition. The axis selecting part 13 is also used to select a
characteristic quantity of which the value is to be calculated.
[0083] The two sets (Parameter 1 and Parameter 2) of parameters
that are thus set for the calculation of a certain characteristic
quantity may be of different kinds or of a same kind. In the latter
case, their contents may be the same or different.
[0084] Let us consider an example wherein Parameter 1 is a set of
data use frequencies N (the number of times data have been used)
and Parameter 2 is a set consisting of combinations of upper and
lower limit values of a band pass filter. Let us further assume
that the characteristic quantity is the maximum amplitude level and
that N-number of largest amplitudes are to be extracted as the
characteristic quantity. In this example, as the waveform data are
read in, the sequential calculating part 21 carries out a band pass
filtering process thereon and then the maximum amplitude level is
calculated. This calculation process is carried out as many times
as the total number of combinations of Parameter 1 and Parameter 2.
The calculation result display part 15 serves to arrange Parameter
1 along the horizontal axis and Parameter 2 along the vertical axis
and to show the evaluation values of the characteristic quantity by
the combinations of the parameters by varying the density, color,
etc. such that the user can easily (visually) grasp the conditions
under which the characteristic shown at a maximum amplitude level
from the waveform data that have been read in such as in which
frequency range and corresponding to which number of used data
parts.
[0085] The process described above for the second embodiment of the
invention is basically the same as shown in FIG. 14 except the
process for calculating the value of the characteristic quantity
becomes as shown in FIG. 17 because there are two sets of
parameters that are set and there is only one characteristic
quantity.
[0086] In this case, Step S3 in the flowchart of FIG. 14 starts, as
shown in FIG. 17, by selecting the first of the parameters in the
set of Parameter 1 (Step S21) and the first of the parameters in
the set of Parameter 2 (Step S22) and calculating the value of the
specified characteristic quantity by using these selected
parameters (Step S23).
[0087] It is thereafter determined whether the calculation in Step
S23 was by using the last parameter in the set of Parameter 2 (Step
S24). If it was not (NO in Step S24), the next parameter in
Parameter 2 is selected (Step S25) and the calculation of Step S23
is repeated. If it was by using the last of the parameters in the
set Parameter 2 (YES in Step S24), it is determined whether the
calculation was by using the last of the parameters in the set of
Parameter 1 (Step S6). If it was not the last (NO in Step S26), the
next parameter in the set Parameter 1 is selected (Step S27) and
Steps S22, S23 and S24 are repeated until the determination in Step
S26 becomes YES and the entire process of FIG. 17, or Step S3 in
the flowchart of FIG. 14, is completed.
[0088] Next, a situation where two sets of parameters (Parameter 1
and Parameter 2) of the same kind are selected is described as the
third embodiment of the invention. An example of such situation
will be where Parameter 1 and Parameter 2 both relate to a
combination of upper and lower limit values of a band pass filter.
In such a situation, the sequential calculating part 21 will read
in waveform data, carry out a specified calculation on the waveform
data that have been read in by using each of the combinations of
upper and lower limit values and then carry out a specified
comparison calculation by using evaluation values obtained based on
Parameter 1 and evaluation values obtained based on Parameter 2
such that the user can clearly grasp which is the frequency range
where the comparison calculation for characteristic quantity brings
about a characteristic most distinctly.
[0089] The process described above for the third embodiment of the
invention is also basically the same as shown in FIG. 14 except
there are two sets of parameters that are selected and there is
only one characteristic quantity. The calculation process by Step
S3 in the flowchart of FIG. 14 is carried out as shown in FIGS. 18
and 19.
[0090] Firstly, a parameter from the set of Parameter 1 is
sequentially called and selected and a specified characteristic
quantity is calculated by using this parameter. This process is
repeated such that the calculation of the value of the
characteristic quantity for given waveform data are completed by
using all of the parameters of the set of Parameter 1 (Steps
S31-S34). Next, similar calculations are done by calling in a
parameter from the set of Parameter 2 (Steps S35-S38). The
parameter set of Parameter 1 and that of Parameter 2 may be
different or the same.
[0091] Next, a preliminarily prepared comparison calculation is
carried out for all combinations of the evaluation values obtained
by using the set of Parameter 1 and those obtained by using the set
of Parameter 2 to obtain a final calculation result (Steps
S39-S45). Explained more in detail, after the first of the
evaluation values obtained by using the set of Parameter 1 is set
as Result 1 (Step S39) and the first of the evaluation values
obtained by using the set of Parameter 2 is set as Result 2 (Step
S40), Result 1 and Result 2 are compared to obtain the final
calculation result (Step S41). If the calculation of Step S41 did
not use the last of the evaluation values of the set of Parameter 2
(NO in Step S42), the next evaluation value is set as Result 2
(Step S43) and Step S41 is repeated. This is continued until the
determination in Step S42 becomes YES and it is then determined
whether or not the last of the calculation results in the set of
Parameter 1 has been used (Step S44). If it was not the last of the
evaluation values (NO in Step S44), the next evaluation value for
the set of Parameter 1 is set as Result 1 (Step S45) and Step S40
is repeated. Thereafter, the comparison calculation of evaluation
values is repeated from the combination with the first of the set
as Result 2 until the determination in Step S44 becomes YES and the
process of Step S3 in the flowchart of FIG. 14 is completed.
[0092] FIG. 20 relates to a fourth embodiment of the invention
according to which characteristic quantities are set along both the
vertical and horizontal axes by the axis selecting part 13 and
either "Waveform 1" or "Waveform 2" is selected by the calculation
mode switching part 12. One waveform data item is obtained by using
the corresponding input indicating area and this waveform data item
is provided to the calculating part 20 through the waveform data
input part 16.
[0093] The calculating part 20 (or its sequential calculating part
21) carries out calculations by using both Characteristic Quantity
1 and Characteristic Quantity 2 that have been given. The
calculations may be carried out as explained above for the first
embodiment of the invention, that is, by calculating the value of
each characteristic quantity of the set of Characteristic Quantity
1 and each characteristic quantity of the set of Characteristic
Quantity 2 and carrying out a preliminarily prepared calculation on
all of the combinations between the evaluation values for the
characteristic quantities of the two sets Characteristic Quantity 1
and Characteristic Quantity 2.
[0094] The calculation result display part 15 serves to arrange
Characteristic Quantity 1 along the horizontal axis and
Characteristic Quantity 2 along the vertical axis and to show the
results of calculation based on combinations of the characteristic
quantities by varying the density, color, etc. as explained above
such that the user can easily (visually) grasp the difference in
the ratio between the values of the characteristic quantities for
given waveform data.
[0095] FIG. 21 relates to a fifth embodiment of the invention. In
each of the embodiments described above, values of characteristic
quantities were calculated based on only one waveform data item.
According to the fifth embodiment of the invention, two waveform
data are transmitted to the calculating part 20 which is provided
with the function of obtaining conditions such as characteristic
quantities appropriate for distinguishing between these two
waveform data.
[0096] Explained more in detail, the waveform data input part 16
provides two waveform data items as objects of comparison such as
"good" products and "no good" products. The calculating part 20
according to this embodiment is provided not only with a sequential
calculating part 21 but also with a comparing part 22. The
sequential calculating part 21 carries out a specified calculation
for each of the waveform data by using all of the combinations of
the characteristic quantities and parameters that have been set.
The comparing part 22 carries out comparison calculations between
those of the evaluation values of the two waveform data obtained by
the sequential calculating part 21, corresponding to the same
combination of characteristic quantity and parameter.
[0097] FIGS. 22 and 23 are a flowchart of operations carried out by
an aid device according to the fifth embodiment of the invention.
To start, the sets of parameters and characteristic quantities set
by the axis selecting part 13 and the axis pattern setting part 14
are read in (Step S51). Next, the two waveform data which are the
object of processing are read in from the waveform data input part
16 (Step S52). The order in which Steps S51 and S52 are carried out
may be reversed.
[0098] Next, values are calculated for combinations of all
parameters and all characteristic quantities that were set by
executing Step S51 for each of these two given waveform data (Step
S53). If the parameters are the pass bands of a band pass filter as
described above, for example, the calculations of values of
characteristic quantities are carried out for the frequency
components that have passed for the waveform data read in by
executing Step S52. This processing of Step S53 is carried out, for
example, by the sequential calculating part 21 executing the
flowchart of FIG. 15 for each of the waveform data. The evaluation
values of the obtained characteristic quantities are transmitted
from the sequential calculating part 21 to the comparing part
22.
[0099] The processing of Step S54 is carried out by the comparing
part 22 executing the flowchart of FIG. 23 in the case where
characteristic quantities and parameters are set on the axes as
shown in FIG. 21. First, the first parameter of the set of
parameters is set to the comparing part 22 (Step S61) and the first
of the set of characteristic quantities is set to the comparing
part (Step S62). Next, comparison calculations are carried out
between the evaluation values obtained for the two waveform data in
the combination of the set parameter and the set characteristic
quantity (Step S63). Different kinds of comparison calculation such
as taking a difference and a ratio may be carried out as long as
the method is appropriate for finding out whether the combination
of the parameter and the characteristic quantity is efficient. If a
difference is used for the comparison calculation, it becomes
easier to find a characteristic with the largest difference by
normalizing the evaluation values.
[0100] When the comparison calculation for one combination of a
parameter and a characteristic quantity is completed, it is
determined whether the comparison related to the last of the set of
characteristic quantities (Step S64). If the calculation has not
been done to the last (NO in Step S64), the next one of the set of
characteristic quantities is set to the comparison part 22 (Step
S65) and Step S63 is repeated. Thus, comparison calculations are
executed with all combinations of a characteristic quantity of the
set of characteristic quantities with a parameter of the set of
parameters and comparison is made between the evaluation values of
two waveform data.
[0101] When the determination in Step S64 becomes YES, it is
determined whether a calculation has been made with the last item
in the set of parameters (Step S66). If the calculations have not
been done to the last item (NO in Step S66), the next parameter is
set to the comparison part 22 (Step S67) and Step S62 is repeated.
Comparison calculations are thereafter continued sequentially for
the next parameter from the first of the set of characteristic
quantities until the determination in Step S66 becomes YES and the
comparison calculation process (Step S54) of FIG. 22 is
completed.
[0102] The comparing part 22 transmits the comparison results
obtained by the execution of Step S54 to the calculation result
display part 15 for a unified display as shown, for example, in
FIG. 11 as a graph showing by density the magnitude of difference
between evaluation values of same combinations of parameter and
characteristic quantity. As explained above, the display may also
be made on a graph with different colors according to the numerical
value such that the result of comparison can be easily grasped by
the user visually.
[0103] In the fifth embodiment of the invention, one set each of
characteristic quantities and parameters is set as in the first
embodiment. Various combinations are possible, however, for the
items to be set as in the second, third and fourth embodiments.
[0104] FIG. 24 relates to a sixth embodiment of the invention
according to which waveform data and parameters are respectively
set along the vertical axis and the horizontal axis by the axis
selecting part 13 and "comparison" is selected by the calculation
mode switching part 12. Thus, specified waveform data are provided
from the waveform data input part 16 to two input indicating areas.
Although calculations are carried out according to the other
embodiments of the invention described above on the basis of only
one waveform data item or two waveform data to be compared,
calculations according to the sixth embodiment are carried out by
using a plurality of waveform data belonging to at least one of the
two groups and evaluation values for this group are examined in a
unified manner.
[0105] For this reason, the calculating part 20 according to the
sixth embodiment of the invention is provided not only with a
sequential calculating part 21 and a comparing part 22 but also
with a representative value calculating part 23. As explained above
with reference to the other embodiments of the invention, the
sequential calculating part 21 serves to obtain evaluation values
for all combinations of specified characteristic quantities and
parameters for given individual waveform data. Thus, if there are a
plurality of waveform data belonging to a single group, evaluation
values are calculated and obtained for each of the waveforms but
the representative value calculating part 23 serves to obtain a
representative evaluation value that represents the group for each
of the combinations of individual characteristic quantities and
parameters and to transmit it to the comparing part 22. At the
comparing part 22, therefore, there are two representative
evaluation values for a same combination of characteristic
quantities and parameters, independent of the number of waveform
data given to the calculating part 20. A comparison calculation is
thus carried out as in each of the other embodiments of the
invention.
[0106] FIGS. 25 and 26 are a flowchart of the processes carried out
according to the sixth embodiment of the invention according to
which the parameter set and the characteristic quantity set that
have been set by the axis selecting part 13 and the axis pattern
setting part 14 are read in (Step S71) and the waveform data
belonging to the two groups that are the objects of processing are
read in from the waveform data input part 16 (Step S72). The order
in which Steps S71 and S72 are carried out may be reversed.
[0107] Next, calculation of values is carried out for all
combinations of the parameters and characteristic quantities set by
carrying out the process of Step S71 (Step S73). If the parameters
are the pass bands of a band pass filter as described above, for
example, the calculations of values of characteristic quantities
are carried out for the frequency components that have passed for
the waveform data read in by executing Step S72. Step S73 of FIG.
25 is carried out by the sequential calculating part 21 according
to the flowchart of FIG. 15 for each of the waveform data. The
evaluation values thus obtained are transmitted to the
representative value calculating part 23.
[0108] Next, the representative value calculating part 23
calculates representative values for all evaluation values
individually for all groups. This is done according to the
flowchart shown in FIG. 26 in the case where characteristic
quantities and parameters are set on the axes as shown in FIG. 24.
First, the first parameter of the set of parameters is set to the
representative value calculating part 23 (Step 81) and the first of
the set of characteristic quantities is set to the representative
value calculating part 23 (Step 82). Next, a representative value
is calculated for the evaluation values obtained for all waveform
data belonging to group 1 in the combination of the set parameter
and the set characteristic quantity (Step S83). Similarly, a
representative value for the evaluation values obtained for all
waveform data belonging to group 2 in the combination of the set
parameter and the set characteristic quantity is calculated (Step
S84).
[0109] The representative value for each evaluation value is
obtained by calculating the average, median, maximum or minimum of
each result obtained for the same combination of characteristic
quantity and parameter for each waveform group. If the evaluation
values of "no good" data tend to be large and those of "good" data
tend to be small, the minimum value may be selected as the
representative value for the group of "no good" data and the
maximum value may be selected as the representative value for the
group of "good" data such that a combination of characteristic
quantity and parameter capable of more dependably distinguishing
between the two groups can be extracted.
[0110] After a representative value is obtained for one combination
of a parameter and a characteristic quantity, it is determined
whether a calculation has been done for the last in the set of
characteristic quantities (Step S85). If the calculation of
representative value has not been done to the last (NO in Step
S85), the next is set to the representative value calculating part
23 (Step S86) and Step S83 is repeated such that representative
values are obtained for combinations of the set parameter and all
of the set of characteristic quantities for evaluation values of
two groups.
[0111] When the determination in Step S85 becomes YES, it is
determined whether calculations have been done to the last of the
set of parameters (Step S87). If the calculations have not been
done to the last (NO in Step S87), the next parameter is set to the
representative value calculating part 23 (Step S88) and Step S82 is
repeated. Thus, representative values of evaluation values are
sequentially obtained from the combination of the next parameter
and the first of the set of characteristic quantities until the
determination in Step S87 becomes YES and Step S74 in the flowchart
of FIG. 25 is completed.
[0112] After the representative values of evaluation values of each
group are obtained, the representative value calculating part 23
transmits these results to the comparing part 22. The comparing
part 22 carried out a comparison calculation process (Step S75) as
explained above with reference to the fifth embodiment and
transmits the results obtained in Step S75 to the calculation
result display part 15 for a unified display (Step S76).
[0113] Since FIG. 24 is based on the first embodiment, it shows a
characteristic quantity and a parameter that are set but it may be
Parameter 1 and Parameter 2 or Characteristic Quantity 1 and
Characteristic Quantity 2 that may be set, as in other embodiments.
One of the groups may be arranged such that only one waveform data
item is inputted. The process of obtaining representative values
may be applied also to situations where characteristic quantities
of only one group are obtained.
[0114] For extracting and adjusting conditions of appropriate
characteristic quantities and parameters for an aid device 10
provided with the function for comparing between two waveform data
or between two groups of waveform data as described above, it is
convenient to proceed according to a flowchart such as shown in
FIG. 27. The flowchart shown in FIG. 27 is for finding suitable
conditions such as characteristic quantities between a group of "no
good" products (NG Group) and a group of "good" products (OK Group)
but it goes without saying that the process shown thereby can be
used for making a distinction between any two different groups.
[0115] To start, combinations of common parameters and
characteristic quantities are used for waveform data of NG Group
and OK Group to calculate values of characteristic quantities,
comparison calculations are carried out between the individual
evaluation values and the results are displayed (Step S91). The
processing of Step S91 may be carried out by operating the device
according to the fifth or sixth embodiment of the invention.
[0116] Next, a specified number of combinations of parameter and
characteristic quantity with significantly different comparison
results are selected from the top (Step S92). Next, these selected
combinations of parameter and characteristic quantity are fixed and
the other parameters are varied to again calculate values of the
characteristic quantities for waveform data of BG Group and OK
Group and the results are displayed (Step S93). For each of these
combinations, another parameter is selected that will maximize the
result of the comparison calculation (Step S94). From the result of
having carried out all these processes until Step S94, the
combination that brings about the largest result of comparison
calculation is selected and accepted as the conditions of
characteristic quantity and parameter for the associated abnormal
sound inspection system (Step S95).
[0117] As an example, let us assume that one of the axes was set to
a characteristic quantity set and the other was set to a parameter
set to carry out comparison calculations and that the following
three combination of characteristic quantity and parameter were
extracted as having high evaluation values: {a, 23}, {b, 27} and
{b, 23}. Next, as Step S93, characteristic quantity "a" and
parameter "23" are fixed and two kinds of parameters A and B
related to characteristic quantity "a" are provided to the
calculating part 20 to obtain evaluation values for all
combinations of parameters A and B for characteristic quantity "a"
by setting parameters A and B respectively along the vertical and
horizontal axes. Of these evaluation values thus obtained, the
largest one is selected (Step S94). Evaluation values are also
obtained similarly for all combinations of parameters A and B for
combinations {b, 27} and {b, 23} (Step S93) and the largest
evaluation value is selected. The characteristic quantity and the
parameter to be actually used by the abnormal sound inspecting
system are selected from these selected candidates.
[0118] In summary, a characteristic quantity and a parameter that
are suitable for distinguishing between waveforms for "good"
products and "no good" products can be extracted by providing
sample waveform data for these products to an aid device of this
invention and a judgment condition for the inspection system can be
determined. Such characteristic quantity, parameter and judgment
condition may be registered in an inspection system for judging the
quality of work pieces produced at the production site, say, by
using an input device manually for the system. Such judgment
conditions created and stored by the aid device may be transferred
to an inspection apparatus by any data transfer process such as
downloading. After the characteristic quantities and parameters to
be used are registered in the inspection system by any of various
methods in addition to the above, judgment processes on work pieces
produced on an actual production line may be started. For example,
a microphone or a vibration sensor may be used to obtain waveform
data on the sound from work pieces and their vibrations may be
obtained and inputted to the inspection system. The waveform data
thus inputted to the inspection system are subjected to the kind of
calculation processes described above and judgments are made as to
whether they are "good" or "no good", the results of such judgment
being subsequently outputted.
* * * * *